Could you explain what you mean? The following article lists Alphabet as a gatekeeper.
https://digital-markets-act.ec.europa.eu/gatekeepers_en
> So if they bring this system in, something which is listening to people real time and using online AI models to translate things, EU might force them to let _any_ 3rd party AI replace it.
If you allow the third party to do that, yes.
> And when someone installs TotallyHonest Co. AI to replace it and there's a massive data leak where they just stored every conversation as-is in an open S3 bucket, who gets the PR flak on HN?
I see this argument often, as often as I hear about leaks. Do you have an instance where Apple was blamed for a leak from a third party? I never heard anybody blaming Apple for Tea app leaks for a recent example, and it is still available on App Store.
Also, an alternative translation app does not have to be provided by a totally random third party vendor. Companies that to me are just as trustworthy as Apple surely will provide alternatives too - Google, OpenAI, Meta, Microsoft or Anthropic.
So I really don't see what's your point here. Don't install the alternatives if you don't trust them.